Programming Cells

Image courtesy of James Brown, Haseloff Lab, University of Cambridge 1

Andrew Phillips Biological Computation Group Microsoft Research

ATGCTTACCGGTACGTTTACGACTACGTAGCTAGCATGCTTACCGGTACGTTTACGAC

Potential

Health Program cells to control tumours

Produce vaccines

Energy Convert CO2 into fuel

Convert sunlight into electricity

2

Challenges

• Programming cells is hugely difficult

– Issues of reliability, toxicity, strain on the host cell

• Systems are highly complex

– Cannot be designed by trial and error

– Requires use of computer software

• Could software for programming cells one day rival software for programming silicon?

3

DNA Structure (A-T,G-C)

4 www.wehi.edu.au/wehi-tv

DNA as a Computing Substrate

• Molecular Scale

– Overcome limits to miniaturisation of silicon chips

– 1GB of information in a millionth of a mm3

– Can self-assemble

– Clean and cheap to manufacture

• Programmable

– Interactions are directly controlled by the sequence

• Massively parallel

Placement and orientation of individual DNA shapes on lithographically patterned surfaces. Nature Nanotechnology 4, 557 - 561 (2009).

DNA Computers Inside Cells

DNA Drugs (Mullis)

Programmable Drugs (Shapiro)

6

Mice + Anthrax

Mice + Anthrax + Aptamer

DNA Strand Displacement

Computation solely by complementary base pairs sticking together (T-A and G-C)

7

Bernard Yurke

DNA Strand Displacement

8

8

DNA Strand Displacement Language

DNA Strand Displacement (DSD) Language

9

Step 1: Program circuit design Step 2: Compile circuit behaviour

Step 5: Insert circuit into cells

Step 3: Simulate circuit

Step 4: Compile circuit to DNA or RNA

Phillips, Cardelli. Royal Society Interface, 2009 Lakin, Youssef, Cardelli, Phillips. Royal Society Interface, 2011

Lakin, Youssef, Polo, Emmott, Phillips. Bioinformatics, 2011

Strand Displacement Logic Gate

Output = Input1 AND Input2

10

Input 1 Input 2

Substrate

Output

Strand Displacement Logic Gate

Output = Input1 AND Input2

11

Input 1

Input 2

Substrate

Output

Strand Displacement Logic Gate

Output = Input1 AND Input2

12

Input 2

Substrate

Input 1 Output

Strand Displacement Logic Gate

Output = Input1 AND Input2

13

Input 2

Substrate

Output Input 1

Strand Displacement Logic Gate

Output = Input1 AND Input2

14

Input 2

Substrate

Input 1

Output

Strand Displacement Logic Gate

15

Strand Displacement Logic Gate

16

Strand Displacement Logic Gate

17

Large-Scale Logic Circuits

18

“In addition to biochemistry laboratory techniques, computer science techniques were essential.” “Computer simulations of seesaw gate circuitry optimized the design and correlated experimental data.”

Turing-Powerful Circuits

Encoding a Stack

Encoding state transitions

19 Lakin, Phillips. DNA Computing, 2011

Model-Checking a DNA Ripple Carry Adder

Localised Circuits

Hairpins tethered to origami

– Increased speed

– Reduced interference

20 Chandran,Gopalkrishnan,Phillips,Reif. DNA Computing, 2011

Programming Living Cells

The “software” of a cell: DNA → RNA → Protein

DNA transcription (real time) Messenger RNA translation

21 Drew Berry www.wehi.edu.au/wehi-tv

Information storage and processing within the cell is more efficient by many orders of magnitude than electronic digital computation, with respect to both information density and energy consumption.

evolutionofcomputing.org

DNA can function across species

A “multi-platform” code

Moving DNA from one organism to another

22

Glowing jellyfish Glowing bacteria

DNA can completely reprogram a cell

23

M. capricolum

M. mycoides

Sequencing

…GTTTCTCCATACCCGTTTTTTTGGGCTAGC…

Synthesis Insertion

Extraction

Transformation

Low-Level DNA Language

A simplified view of DNA instructions

c0040 b0015b0034

r0040

ATGCTTACC GGTACGTT TACGACTACGT AGCTAGCAT

GGUACGUU UACGACUACGU

Protein

polymerase

ribosome

Protein

24

Promoter (start) Terminator (stop)

Protein Coding Region Ribosome Binding Site

High-Level DNA Language

Given a design, automatically determine the DNA

????????? ????????? ????????????? ?????????

25

Genetic Engineering of Cells (GEC)

A language for programming cells

26

Step 1: Program device design

Pedersen & Phillips. Royal Society Interface, 2009

Step 2: Compile device behaviour Step 3: Simulate device

Step 5: Insert DNA into cells Step 4: Compile device to DNA

With Michael Pedersen, Matthew Lakin

Programming a receiver device

Signal crosses cell wall and binds to Receiver

27

Programming a receiver device

28

Programming a receiver device

29

Characterising a receiver device

Model Simulation

Experimental Data

30

Spatial Receiver Device

Model Simulation Experimental Data

31

Programming Turing Patterns

Cells that communicate to perform complex functions

With Neil Dalchau, James Brown, Jim Haseloff Service. Science, 2011

Modelling Biophysics: Thresholding

GEC

Gibthon CellModeller

DB

DB

Image Processing

pA B pB A

pS

S

B

With Tim Rudge, James Brown, Jim Haseloff

S (AHL)

Modelling Biophysics: Thresholding

MHC class I Antigen Presentation

©Diego Accorsi 2011. A Master’s Research Project submitted in conformity with the requirements for the degree of Master of Science in Biomedical Communications (MScBMC), Faculty of Medicine, University of Toronto.

Peptide Optimisation

• Peptide-MHC complexes generally need to be stable for many hours or days at the cell surface

• Peptide optimisation: high affinity peptides are preferentially selected for presentation

• Optimisation in the ER is typically limited to tens of minutes

– How is such high optimisation is achieved in so little time?

– What are the main mechanisms of optimisation?

36

MHC Class I Model

37 Phillips, Dalchau et al. PLoS Computational Biology, 2011

Automatic generation of reactions

38 Phillips, Dalchau et al. PLoS Computational Biology, 2011

Automatic generation of ODEs

Phillips, Dalchau et al. PLoS Computational Biology, 2011

Steady-state ODE analysis

Steady-state experiments

Tapasin enhances peptide presentation by 1/ui

Measured off-rates ui for SIINFEKL peptides

Phillips, Dalchau et al. PLoS Computational Biology, 2011

Delayed egress vs. Enhanced off-rate

Measured transit time of MHC to cell surface

42

Fast transit => fast optimisation => enhanced off-rate

Phillips, Dalchau et al. PLoS Computational Biology, 2011

Time-dependent experiments

Representative peptides Plow, Pmed, Phigh

Phillips, Dalchau et al. PLoS Computational Biology, 2011

Identify differences between alleles

Allele-specific peptide binding

44

B4402 B2705

Phillips, Dalchau et al. PLoS Computational Biology, 2011

Optimisation of HIV peptides

45 Off-rates and abundance Cell-surface presentation

Kinetic model

BIMAS Off-rate predictor Viral protein sequence

MGARASVLSGGELDRWEKIRLRPGGKKKYK....

Peptide off-rate (s-1) Peptide off-rate (s-1)

Biological Computation Group

46

Immunology Developmental Biology Synthetic Biology DNA Computing

Organism Organ Cell Molecule

SPiM: Processes SBT: Stem Cells GEC: Genetic Devices DSD: DNA Circuits

Biological Modelling Engine: A common language runtime for Biological Computation

Acknowledgements

47

Luca Cardelli

James Brown

Michael Pedersen

DNA Computing Synthetic Biology (Cambridge)

Neil Dalchau

Matthew Lakin

Filippo Polo

Simon Youssef

Jim Haseloff

Tim Rudge

Joern Werner

Tim Elliott

(Southampton)

Mark Howarth (Oxford)

Immunology

Leonard Goldstein

(Cambridge)

Microsoft Research

Stephen Emmott

48